In this paper, we propose a density control algorithm for wireless sensor networks to keep as few as possible sensors in active state to achieve a connected coverage of a specific area of interest. Inactive sensors can turn off sensing modules to save energy for prolonging the network lifetime. Unlike other algorithms, the proposed one does not rely on position information or ranging information of sensors. It just requires each active sensor to periodically send two beacons of different transmission ranges. Sensors can then decide to stay active or inactive according to received beacons. When any active sensor runs out of energy or fails, one or more inactive sensors can switch to the active state to take over the surveillance responsibility. The proposed algorithm is thus of the capability of fault-tolerance. Under the assumption of sufficiently high density of sensors and the assumption of R C ≥ 2RS, the algorithm can approximate the optimal connected coverage, where RC and RS are the radio communication radius and the sensing radius of sensors, respectively. Furthermore, we perform simulation experiments to investigate the impact of the sensor density and the ratio α on algorithm performances, where α, (1/√3) < α < 1, is the ratio of the transmission ranges of the two beacons.